Download PDF

The Australian Journal of Agricultural and Resource Economics, 50, pp. 85–101
Price or politics? An investigation of the causes
of expropriation*
Oxford,
The
AJAR
&ãgr;
1364-985X
1ORIGINAL
50
Price
R.
Duncan
Australian
or
Australian
UK
politics
ARTICLE
Journal
Agricultural
of Agricultural
and Resource
and Resource
Economics
Economics
Society Inc. and Blackwell Publishers Ltd 2006
Blackwell
Publishing
Ltd
Roderick Duncan†
Expropriations of foreign direct investment in developing countries are typically
blamed on political and economic crises in those countries. Developing a new database of expropriations in the minerals sectors of developing country exporters, I show
that expropriations were correlated with mineral price booms and that democratic
governments were more likely to expropriate. No link is found between expropriations
and political or economic crises, except at independence. A better explanation of
expropriation would be opportunistic behaviour by host governments when profits of
investments are high. In two developed countries, Australia and Canada, expropriations are also found to occur during price booms.
Key words: expropriation, foreign direct investment, natural resources.
1. Introduction
Between 1967 and 1974, governments in many major copper-exporting developing economies wholly or partially nationalised the foreign-owned copper
mines in their countries. These countries included Chile, Peru, Zaire and
Zambia. A similar wave of expropriations occurred among the major bauxiteexporting developing countries during 1974 and 1975, including the Dominican
Republic, Guinea, Guyana, Jamaica and Suriname. Since the 1980s, there has
been no act of expropriation among the same copper and bauxite exporters.
Expropriation here means any act by which a government seizes a greater
share of an investment project than it was entitled to under the contract with
the foreign investor. The meaning of expropriation will be expanded upon
below.
What caused these sudden episodes of expropriation? Why were expropriations occurring in different countries at the same time? And why did these
expropriations stop occurring? This article provides evidence that these
expropriations were caused by the high real prices of copper and bauxite and
that real output prices are a strong predictor of expropriation across the minerals
sectors in both developing and developed countries.
*
The author thanks Anne Krueger, Aaron Tornell and participants at the Stanford University Trade and Development Workshop.
†
Roderick Duncan (email: [email protected]) is a Lecturer in the School of Marketing
and Management, Charles Sturt University, Bathurst NSW 2795, Australia.
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
doi: 10.1111/j.1467-8489.2006.00316.x
86
R. Duncan
Before the late 1960s, real prices of minerals had been on a downward
trend for nearly a century. The downward trend had created an environment
of pessimism around the production of minerals in developing countries, well
illustrated in the words of H. W. Singer (1950, p. 577) at the American Economics Association meeting in 1950:
It is a matter of historical fact that ever since the seventies [1870s] the
trend of prices has been heavily against sellers of food and raw materials
in favour of sellers of manufactured articles.
This environment dramatically changed during the mineral price booms
of the late 1960s and early-mid-1970s. Price booms created an expectation of
higher future prices and imminent worldwide shortages. Paul Ehrlich’s Population Bomb appeared in 1968 and forecasted food riots in the United States
by the 1980s. The Club of Rome published Limits to Growth in 1972 forecasting
mineral shortages in the near future.
In the new environment of high and (predicted) rising prices, investment
contracts based on export taxes appeared to developing countries as very bad
deals, since investment profits rose to a much greater extent than did export
tax revenues. Developing countries remedied this situation by expropriating
existing investment and by switching to contracts based on profit-sharing.
Prain (1975) estimated that copper exports in which the governments held
any equity comprised only 2.5 per cent of world exports in 1960, but by 1970
this share rose to 43 per cent. By 1970, Prain (1975, p. 223) notes:
[m]ore than a quarter of the world’s copper was being produced by
mines totally owned by the government, 12 per cent by companies in
which the government had a majority interest, and 5 per cent in which
the government had minority interests.
The international environment underwent another change at the end of the
1970s. The minerals sectors entered a price slump that lasted throughout the
1980s and 1990s with a few exceptions. The expected high prices and worldwide shortages never eventuated. In these decades, there was no incentive to
expropriate for host countries as mine profitability was low.
2. Definition of expropriation
In this article expropriation is defined as any act by which a country gains a
greater share in the output of an investment than it was entitled to under the
original contract between the foreign investor and the country’s government.
Expropriation includes:
• seizure of capital, including mining equipment, reserves of the mine, or
mining rights (a complete seizure of domestic assets of the foreign company is known as ‘nationalisation’)
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
Price or politics
87
• compelled sale of mining company shares to the government or domestic
nationals; or
• raising taxes on company revenues or profits
where this act was not a part of the original agreement between the foreign
investor and the government. To qualify as expropriation, the tax increases in
the third form of expropriation must be aimed at the investment, rather than
economy-wide increases.
It is difficult to determine which of these forms of expropriation is more
common and how large the total level of expropriations has been, as existing
data on expropriations of foreign direct investment is incomplete. Williams
(1975) compiled data on nationalisations of foreign investment, only a subset
of all expropriations. For the period 1956–1972, Williams found that the value
of assets seized by developing country governments was US#10 billion (in
1972 values) or nearly 25 per cent of the total stock of foreign direct investment at the end of 1972. The level of total expropriations would obviously be
a higher percentage of total foreign investment.
3. Three models of expropriation
This section presents three competing models of expropriation. From these
models a regression model is developed to test the predictions of each.
3.1 The contract model of expropriation
Models of foreign direct investment (FDI) in the economics published works
that have considered the possibility of expropriation tend to fall into two
camps. The incentive-compatible contract papers such as Atkeson (1991),
Thomas and Worrall (1994), and more recently Schnitzer (1999, 2002) develop
contracts which are expropriation-proof. The random expropriation papers
such as Eaton and Gersovitz (1984), Cole and English (1991), Raff (1992)
and Veugelers (1993) consider equilibria in which expropriation occurs with
a positive probability. The model of Cole and English will be used here.
Cole and English (1991) developed a model of expropriation of FDI in a
country that is subject to stochastic shocks to production. Using the Cole
and English notation but replacing their stochastic variable zt with the output
price of the FDI, Pt, the production function in per capita terms of an FDI
in a host country is:
yt = Pt f (kt)
(1)
where yt and kt are the per capita level of output and capital in the FDI.
Assuming that host countries maximise the present value of expected utility,
Cole and English modelled host country’s expropriation decisions as trading
off the present value of extra revenues gained today and in the future from
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
88
R. Duncan
expropriation to the loss of future reinvestment and so productivity, assuming that foreign investors would not reinvest after expropriation.
Cole and English (1991) then defined the expropriation set which is the
set of values for Pt for which the expected present value of expropriation
exceeds the expected present value of not expropriating the FDI. Assuming a
case of a constant relative risk aversion utility, where the utility of present
consumption, ct, is:
u (ct ) =
ct1−γ − 1
1−γ
(2)
Cole and English showed that there are three cases for the expropriation set,
corresponding to whether the risk aversion parameter, γ, is greater than one,
less than one, or equal to one.
In the case where γ is less than one, the host country will expropriate when
output prices are high. They call this the ‘opportunistic’ explanation of
expropriation. This was also the prediction of the model in Section VI of
Eaton and Gersovitz (1984) and in Raff (1992). In the case where γ is greater
than one, the host country will expropriate when output prices are low. Cole
and English (1991) call this the ‘desperation’ explanation of expropriation. In
the case where γ is equal to one, utility is logarithmic, and expropriation either
always occurs or never occurs. The third case is not considered as expropriation of FDI and seems to occur only occasionally.
Probabilistic interpretations of the two cases in Cole and English (1991;
‘the contract model’) are as follows: (i) opportunistic explanation: the
probability of an expropriation increases in years in which the real price of
the output mineral is high (‘boom years’); or (ii) desperation explanation: the
probability of expropriation increases in years in which the real price of the
output mineral is low (‘bust years’).
3.2 The political revolution model of expropriation
This model of expropriation states that expropriations follow on from and
are a result of political revolutions in a country, typically a former colony.
Kobrin (1980, p. 74) labelled these expropriations ‘mass expropriations’ and
described them as ‘typically associated with sweeping and violent upheavals,
which transformed the basic governmental structure and political-economic
ideologies of the nation involved.’
Typically, the political upheavals that led to this change in the balance
of power between the colony and its master were radical, and it was these
changes that were widely believed to be the principal cause of expropriation.
A survey of executives of 233 US companies in the 1970s gave the most commonly cited explanation for government expropriation to be, from Basche
(1979, p. 8), ‘[p]olitical change involving not only a change of political leaders and
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
Price or politics
89
government personnel – but also, more fundamentally, a change in political
and social ideology.’
If this explanation were the true model of expropriations, one would expect
that expropriations occur at the same time or shortly after radical changes in
political and social structures in a developing country. In probabilistic terms
this model yields the hypothesis that the probability of expropriation
increases if there was a political crisis involving a large change in the political
structure of that country in that year.
3.3 The political risk model of expropriation
Published work exists, mainly within business school journals, which seeks to
explain expropriatory actions by governments on the basis of the macroeconomic and political environments in those countries. This theory draws from
both politics and international business and attempts to calculate what it
terms ‘political risk’ to international business. A review of this published
work is contained in Kobrin (1979).
In the political risk published work, expropriatory behaviour is driven by
political pressure on the domestic country government. However, the political
risk model proceeds further than the political revolution by investigating the
forces behind political pressures to expropriate.
The explanation behind this model is that under criticism from either
opposing political forces, the media and/or popular forces, host governments
expropriate foreign investments. An expropriation could have a symbolic
value, using foreign investors as a scapegoat for domestic troubles, or an
expropriation might have a real effect, by raising government tax revenues or
reducing the balance of payments.
One such source of political pressure might be poor economic performance within the domestic economy. Another source of political pressure
might be the balance of payments. In probabilistic terms this model yields the
hypotheses that: (i) the probability of expropriation increases if domestic real
gross domestic product (GDP) growth was low that year; and (ii) the probability of expropriation increases/decreases if the domestic balance of payments
as a percentage of GDP was worse that year.
3.4 Review of the empirical published work
The empirical published work on expropriation of foreign investment is not
well integrated into the theoretical published work. This failure stems from
the fact that the empirical published work comes out of work in the political
revolution and political risk areas that have no formal model at their foundations. The existing empirical published work is then a set of regressions where
the choice of variables included is based on whether a plausible argument can
be made that the variables might be expected to influence expropriations.
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
90
R. Duncan
The political revolution model of expropriation has not been truly tested.
However, Jones (1980) found some support for a relationship between
domestic unrest and expropriation, whereas Jodice (1980) did not. The political risk model has been directly tested with Knudsen (1974), Jones (1984),
and Picht and Stüven (1991) finding a negative correlation between real GDP
growth and the probability of expropriation. However, Jodice (1980) and
Picht and Stüven (1991) draw opposing signs for estimates of the relationship
between balance of payments difficulties and expropriations. The results of
the current published work, on the whole, have been weak.
4. The data and the regression model
In order to test these three models of expropriation, a database of government
expropriations for seven major minerals from 1960 to 2002 was constructed.
This database is available on the author’s website. The minerals selected were
bauxite, copper, lead, nickel, silver, tin and zinc. These minerals were selected
because they are important minerals for developing countries’ export revenues
and for developed countries’ industry.
The countries in the sample are the eight largest developing country
exporters for the period 1965–1975 for each of these minerals. (Former)
Communist countries were excluded as these countries were not sufficiently
linked to the world mineral markets over this period.
Any seizure of assets, demand for equity stake or increase in taxes by the host
government that was not a condition of the original contract was considered
to be an expropriation. Information on host government activities in the minerals
sector was gained from various issues of US Bureau of Mines Minerals Yearbook.
An observation in this database is whether an expropriation took place in a
foreign-financed project for that mineral, in that country, in that year.
Out of the total of 2408 observations of commodity–country–years, there
were only 50 cases of expropriation. Out of these 50 cases, 20 were acts of seizure of company assets, 12 were uncontracted increases in the equity share
held by the host government, and 18 were uncontracted increases in taxes on
the mining companies. Two events that were government-mandated increases
in workers’ wages were classified as tax increases, as these effectively raised
company costs much as a new tax. Separate regressions were run for each
form of expropriation, but the results were not qualitatively different, so only
the pooled regressions are presented. Of the 27 countries included in the
database, expropriations occurred at least once in 18 of the countries.
Table 1 sets out these acts of expropriation by mineral and by decade, as well
as average real prices of the seven minerals. Real prices of minerals were obtained
from the United States Geological Survey ‘Historical Statistics’ website. These
real mineral prices were then normalised to have a mean of one over the period
1960–2002. All subsequent references to ‘price’ are the normalised real price.
For the purposes of testing the contract model, a definition of boom years
and bust years is needed – years when a real mineral price was significantly
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
Price or politics
91
Table 1 Number of expropriations in selected country/industries by mineral and by decade
(number of expropriations)
Mineral
1960 –1969
Bauxite
Copper
Lead
Nickel
Silver
Tin
Zinc
All minerals
Ave. minerals price
1
5
1
0
1
2
2
12
1.03
1970 –1979
15
6
1
0
2
11
2
37
1.32
1980–1989
1990–2002
All years
0
0
0
1
0
0
0
1
1.09
0
0
0
0
0
0
0
0
0.67
16
11
2
1
3
13
4
50
1
above or significantly below its long-term average price. None of the empirical
results depended on the critical value chosen, so values 66 per cent and 150
per cent of the long-term average price were arbitrarily used as the definition
of bust and boom years.
Three different measures of ‘political crisis’ were constructed. Out of the 27
developing countries in the database, 15 became independent after 1955. A
dummy variable called ‘Newly Independent’ is used to indicate whether that
country became independent recently. Often, gaining independence can be a
lengthy process, so an immediate correlation between independence and
expropriation might not be expected. In this dataset, the largest level of economic
and statistical significance was found to be within 5 years of gaining independence.
A second measure of political crisis is a large change in the institutions of
the country. A dummy variable called ‘Constitutional Crisis’ is used to indicate that change. The data on constitutional crises is derived from the Polity
IV database described in Marshall and Jaggers (2001).
The Polity IV database contains indices for the level of democracy and
autocracy of the political institutions for each country–year. A combined
measure for each country–year is derived by subtracting the autocracy score
from the democracy score to generate a scale of −10 (least democratic) to 10
(most democratic) for the country’s institutions in that year. This index is
labelled ‘Democracy’. The Democracy index coverage does not include the
countries of New Caledonia and Suriname, and nor does it include countries
in years in which those countries were colonies. A constitutional crisis in any
country–year is indicated by a change in this combined index of three points
or more from that of the previous year, according to Marshall and Jaggers
(2001). Smaller thresholds for crises and lagged crises were examined, but did
not significantly affect the analysis.
A third measure of political crisis is whether there was a change in the
chief executive in that country in that year. A dummy variable called ‘Change
in Government’ is used to indicate a change in the chief executive. Data on
changes in executives was obtained from Beck et al. (2001) and from Banks
et al. (2003) to narrow the change down to a particular year.
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
92
R. Duncan
Table 2 Sample statistics
Expropriation
Constitutional crisis
Change in government
Growth of real GDP
Balance of payments
Democracy index
Mean
Overall
SD
Min.
Max.
Between-country
SD
No. of
observations
0.021
0.12
0.17
3.65
−3.26
−0.63
0.14
0.32
0.38
5.56
5.74
6.79
0
0
0
−48.8
−37.1
−9
1
1
1
67.9
37.6
10
0.03
0.09
0.1
2.01
3.48
5.08
2408
2211
2408
2292
1977
2104
SD, standard deviation; GDP, gross domestic product.
Macroeconomic data on the real growth of GDP and the balance of payments
as a percentage of GDP are taken from the United Nations Statistics Division
(2005) ‘National Accounts’ website and from the International Monetary
Fund’s (various years) International Financial Statistics, respectively. Growth
of real GDP is reported as percentages. The balance of payments is normalised by presenting it as a percentage of nominal GDP. Sample statistics for
the data are set out in Table 2.
A regression model can be based around the three models of expropriation. The contract model would suggest that expropriation occurs when the
real output price of the project either is greater than or less than some critical
value, D, which includes any punishment imposed on the country due to the
expropriation.
Assume this equation is a linear function of price, an expropriation is
observed in industry j and country i in year t if and only if
αijt + βijt pjt > Dijt
(3)
where βijt will be positive in the opportunistic explanation of expropriation and
negative in the desperation explanation. To increase the power of the regressions,
it is further assumed that the αijt’s and βijt’s are the same across countries and
industries and time.
Dijt includes variables that could otherwise influence expropriation, such as
the growth rate of real GDP or the occurrence of political crises. Suppose
one can linearise D, so that
Dijt = γδijt.
(4)
Then expropriation is observed when
α + β pijt − γδijt + εijt > 0
(5)
where εijt is assumed to have a normal distribution with mean zero and variance
one. It is this equation that is estimated here using probit analysis. This is called
the ‘Expropriation Equation’.
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
Price or politics
93
5. Results
5.1 Evidence from summary statistics
Table 3 presents some sample statistics of the expropriation variable. Using
the definition of boom year discussed in the previous section, the contract
model predicts that the conditional probability of expropriation in a boom
year is higher than the conditional probability of expropriation in a non-boom
year. Only lead and zinc do not follow this pattern.
From all seven minerals a clear pattern arises. Although boom years
constituted only 7 per cent of the total years observed, 34 per cent of the
expropriations occurred in these years. Overall, a default was seven times more
likely to occur in a boom year than it was in a non-boom year. An F-test that
the probabilities of expropriation are the same inside a boom year as a nonboom year gives a test statistic of F(1, 2406) = 58.8. This hypothesis that
there is no relationship between minerals price booms and expropriation is
rejected at far below the 1 per cent confidence level. The dataset sample statistics
do support the opportunistic interpretation of expropriation.
The desperation explanation suggested that there might be a link between
low prices for minerals and expropriations. For various measures of a ‘bust’
year at 50 per cent, 66 per cent or 75 per cent of average real prices, the probability of expropriation is far lower for bust years than for non-bust years.
The desperation interpretation of expropriation is not pursued any further, as
the dataset does not support it.
If the political revolution model were correct, the probability of expropriation should be higher in years of political crisis. No statistically significant
effect was found within 1 or 3 years after independence; however, the probability of expropriation is statistically different at the 1 per cent level within 5
years of gaining independence, with a test statistic of F(1, 2406) = 12.9. For a
constitutional change and a change in executive government, the probability
of expropriation coinciding with years of crisis is not statistically significant
at any level, with test statistics of F(1, 2046) = 0.04 and 0.34, respectively.
This finding is robust for several different time lags.
Table 3 Probability of an act of expropriation in selected country/industries by mineral (percentage of observations in which an expropriation occurred)
Bauxite
Copper
Lead
Nickel
Silver
Tin
Zinc
All minerals
Unconditional
In boom year
Not in boom year
4.7
3.2
0.6
0.3
0.9
3.8
1.2
2.1
50.0
25.0
0.0
6.3
2.5
14.5
0.0
10.1
3.6
2.1
0.6
0.0
0.7
2.0
1.3
1.5
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
94
R. Duncan
5.2 Results from probit regressions
Tables 4 and 5 present the results of probit regressions on the Expropriation
Equation set up in section 4. The dependent variable is the incidence of
expropriation in a particular mineral, country and year.
Table 4 tests whether the results are robust to panel data correction methods.
The first column of Table 4 presents the results for a standard probit regression. Panel data corrections for probit models are problematic. No consistent
fixed-effects probit estimation method exists (see Baltagi 1995, p. 181), so
country-specific effects are corrected in the second regression in Table 4 through
including a dummy variable for each country individually. The third column
presents the results from a random-effects probit regression.
The results of the regressions for price booms are robust across the panel
data corrections. No correlation is found between years of constitutional
change or of executive change and expropriation, whereas there is some evidence of a positive effect between gaining independence within 5 years and
expropriation. If the individual features of each country are corrected with a
dummy variable as in Regression 2, the Newly Independent variable is found
to be statistically insignificant, whereas a positive and statistically significant
effect is found for the Balance of Payments. Because of the problems with the
fixed-effects estimation and with interpretation of the random-effects model,
Regression 1 is chosen as the reference model, but the Balance of Payments
Table 4 Results of the probit regressions for the expropriation equation (dependent variable
is the probability of expropriation)
Regression:
Price boom dummy
Newly independent
Constitutional crisis
Change in government
Growth of real GDP
Balance of payments
Constant
Observations
Groups
Pseudo R2
Log likelihood
(1)
Standard
probit
(2)
Unconditional
fixed effects
panel probit
0.992***
(0.165)
0.94***
(0.252)
0.067
(0.249)
−0.066
(0.201)
−0.004
(0.012)
0.008
(0.011)
−2.25
1906
–
0.114
−168.7
1.18***
(0.199)
0.389
(0.357)
0.033
(0.274)
−0.002
(0.229)
0.012
(0.017)
0.027**
(0.013)
–
1115
14
0.199
−135.5
(3)
Random effects
panel probit
1.09***
(0.184)
0.626*
(0.32)
0.072
(0.262)
−0.022
(0.217)
0.004
(0.014)
0.018
(0.013)
−2.4
1906
24
–
−161
Note: ***, **, and * denote statistical significance at the 1, 5, and 10 per cent levels, respectively. Standard
errors are in parentheses.
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
Price or politics
95
Table 5 Probit model with interaction of real price and democracy variables (dependent variable is the probability of expropriation)
Regression:
Price boom
Decade: 1960s or 1970s
Newly independent
Balance of payments
Democracy
Democracy and price
Interaction
Constant
Observations
Pseudo R2
Log likelihood
(4)
Standard probit
(5)
Standard probit
(6)
Standard probit
0.99***
(0.19)
1.68***
(0.37)
0.45
(0.27)
0.011
(0.012)
0.048***
(0.01)
–
1.07***
(0.20)
1.74***
(0.41)
0.46*
(0.27)
0.01
(0.013)
–
1.00***
−0.27
1.42***
(0.45)
–
0.039***
(0.01)
−3.58
1892
0.253
−136.1
0.033*
(0.017)
−3.31
1075
0.18
−68.2
−3.52
1895
0.246
−137.5
0
(0.036)
–
Note: ***, **, and * denote statistical significance at the 1, 5, and 10 per cent levels, respectively. Standard
errors are in parentheses.
variable is retained as it does have explanatory power in one form of the
regression.
One potential problem with the Newly Independent variable is that 11 of
the countries in the sample became independent in the 1960s and 1970s, and
so the issue of independence may become confused with worldwide factors
special to those decades. To try to capture the effects of worldwide factors in
the 1960s and 1970s, I introduce a dummy for those decades.
Table 5 explores the form of Regression 1 with several variants. Regression
4 in Table 5 introduces decade dummies for the 1960s and 1970s and tests for
the effect of democracy. An interaction variable between real minerals price and
the level of democracy of a country is used in Regression 5. In Regression 6,
the dataset is restricted to only those countries that were independent before
World War II, in order to determine whether the price boom effect is unrelated
to questions of independence.
The coefficient of price booms and the decade effect of the 1960s and 1970s
are positive, statistically significant and robust across the specifications. Adding the democracy variables reduced the explanatory power of the independence variable. Achieving independence up to 5 years prior did have a positive
effect, but it was only statistically significant at the 10 per cent level once the
interaction effects of democracy are taken into account and was not significant
if the democracy variable is included. The coefficient of the balance of payments
does not have a statistically significant effect across any of the variants. The
coefficient of democracy and the coefficient of interaction of democracy and
real price are positive and statistically significant at the 1 per cent level. All of
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
96
R. Duncan
the results are robust to the restricted dataset of those countries that were
independent before World War II.
5.3 The significance of price and contract form
The probit regressions explored the relationship between the output price
of a foreign direct investment and the behaviour of the developing country
government in which the investment operates. This revealed a strong positive
relationship between the price and the probability of expropriation by the
government. As the period of the 1960s and the 1970s saw many more price
booms than the period of the 1980s and 1990s, this can partly explain why
there were so many more expropriations during the early period. The lone
recorded expropriation in the 1980s and 1990s occurred in the nickel sector
in the Dominican Republic in 1987–1988 when the real price of nickel almost
tripled the previous year’s average.
There is strong support for the opportunistic explanation of expropriation
in this dataset. Comparing the predicted probabilities of expropriation from
Regression 5 in Table 6, price booms had a large effect on the probability of
expropriation in the early period and a lesser impact in the later period.
There has also been a distinct change in the behaviour of developing country
governments from the early period, as revealed by the large and significant
coefficients on the decade dummy. Even taking into account the lower prices
in the late period, governments are now less likely to expropriate than in the
early period.
Given a tendency to expropriate in times of high prices or profitability, the
most stable form of FDI contract may then be a pure ‘resource rent tax’ contract similar to that proposed by Garnaut and Clunies Ross (1975). Under
this contract government tax revenues are derived solely from a profits tax
where the rate of tax of profits may even be progressive in the level of profits.
By granting a large portion of the revenue stream of the project to the government in the good states, this contract lessens the incentive to expropriate. The
cost is that the developing country government bears a large part of the price
and output risk of the contract, which the government may have problems
hedging against.
Extreme cases of these new contracts are the ‘service contracts’ some
mineral companies have signed with developing countries. Under these
contracts, mining companies explore, develop and even produce from an ore
body for a flat fee from the host government. The host government owns all
the capital in the project and bears all of the price and output risk of the
project. These projects bear no risk of expropriation.
5.4 The significance of politics and political crises
The political revolution model of expropriation is strongly rejected by these
data, except to the extent that gaining independence is found to have had a
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
Price or politics
97
Table 6 Predicted probability of expropriation using the estimated coefficients from the
expropriation model (predicted probability in per cent)
Using the estimated coefficients from Regression (1)
Price boom
No
Yes
Newly independent
No
Yes
1.2
9.5
10.4
37.5
Using the estimated coefficients from Regression (5)
Not newly independent
Price boom
No
Yes
−10
1.3
6.1
Decade:
1960s and 1970s
Decade: 1980s, 1990s
and 2000–2002
Democracy index
Democracy index
0
3.3
22.1
10
7.4
50.4
−10
0.0
0.1
0
0.0
0.6
10
0.1
4.2
Newly independent
Price boom
No
Yes
−10
3.8
13.8
Decade:
1960s and 1970s
Decade: 1980s, 1990s
and 2000–2002
Democracy index
Democracy index
0
8.4
37.8
10
16.1
68.1
−10
0.0
0.2
0
0.1
2.0
10
0.3
10.2
Note: All other variables held at their period averages. Assumed value of real price during boom years is
2 and during non-boom years is 1.
significant effect. Expropriations do not necessarily coincide with or follow
from significant political changes or from periods of internal war or civil
strife. Political language of ‘sovereignty’ and ‘revolution’ may thus just be
cover for actions motivated by more mundane considerations.
The results of the regressions in Table 4 suggested that the fact of gaining
independence was a significant contributor to expropriation. However, in
Table 5, once account is taken of the worldwide factors in the 1960s and 1970s
and account is taken of the contribution of democracy to expropriation, the
economic and statistical significance of independence falls away.
The importance of the democratic nature of the country’s institutions
stands in stark contrast to the irrelevance of constitutional crises within
those institutions. The more democratic the institutions of the country, the
more likely it was to expropriate, as shown in Table 6. Table 5 suggests that
this relationship between expropriations and democracy comes through the
interaction between price and democracy. The more democratic the political
institutions of the host country are, the more sensitive to price is the decision
to expropriate.
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
98
R. Duncan
Yet the data on political change did not reveal any link between change in
executive and expropriation, thus expropriation is not caused by turnover
of political power. This fact supports the idea that it is the higher potential
for turnover among political interest groups in a democracy that leads to
a higher discount rate among those groups and thus leads to a higher
tendency to expropriate in those countries, rather than the actual turnovers
of power.
A different explanation might be that contained in Tornell and Lane
(1999). One interpretation of their model is that interest groups take turns
ruling a country. These interest groups all derive benefits from a common
resource, but the party in power has the option of taking the whole resource
for itself. If the interest group currently in power does not trust future interest
groups not to expropriate when they come into power, the current interest
group will expropriate now rather than wait and let a future interest group
expropriate. Democracies, in this case, would be very prone to have governments
that expropriate.
These results indicate a negative relationship between political freedom
and security of property rights, at least for foreign investors. Barro and Lee
(1994) found a positive relationship between political freedom and growth of
real GDP for a group of 116 countries from 1965 to 1985. The traditional
explanation for this result would be that democracy restricts the actions of
government and thus promotes property rights. Bohn and Deacon (2000) used
this argument when they discovered that democracy and aggregate investment
were positively correlated. However, the results in this article suggest that
political freedom and property rights may not be positively related. Democracy
may lead to less stable property rights, particularly for foreign investors.
5.5 The (in)significance of macroeconomic factors
The macroeconomic variables from the political risk model appear to have
little predictive power in this dataset. Neither real GDP growth nor balance
of payments have a statistically significant relationship to expropriation at
even the 10 per cent level of confidence, except in the problematic fixed effect
probit regression in Table 4. There is no support for the political risk model
of expropriation in this dataset.
6. Are developed countries immune to this pressure?
No features of the contract or political risk models are predicated on the host
country being a developing economy. The political pressures that were mentioned as a possible explanation for expropriations are not particular to
developing countries alone. In fact, the democratic nature of political institutions in most developed countries may make them more prone to expropriation
if the link between democracy and expropriation uncovered in the previous
section is correct.
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
Price or politics
99
However, past empirical studies, including all of the articles summarised in
section 3, consider only expropriations that occurred in developing countries.
Is this focus justified?
Australia and Canada are two developed economies that are also large
minerals exporters. Searching through various issues of the Minerals Yearbook in the same seven minerals extraction sectors as were examined in the
developing economies, 11 acts of expropriation either by state or federal governments in Australia and Canada were found between 1960 and 1996.
The conditional probability of expropriation given a boom year is 9.3 per
cent, whereas it is 1.33 per cent given a non-boom year. An F-test for the
equality of the probabilities of expropriation for boom years and non-boom
years is F1,592 = 16.0 and so rejected at 1 per cent significance. Australia and
Canada were more likely to expropriate the value of foreign direct investments
during periods of high real prices for the output of the projects. This suggests
that developed countries do not behave very differently from developing
countries in this respect.
An interesting point to note is the role that the federal structure of the
Australian and Canadian constitutions may have played in the expropriations.
Of the 11 expropriations reported, four were by state or provincial governments. In one case, that of the Canadian copper industry, the Canadian federal
government reacted to a rise in provincial taxes by introducing a limited
federal tax write-off of provincial taxes.
The federal nature of the Australian and Canadian constitutions grants
much wider powers to state governments than does a constitution styled on
that of the United States. These broader powers of taxation mean that in the
Australian and Canadian cases, there are two independent governments that
can expropriate the value of the same foreign investment. The foreign investment is then a common resource of the two governments, which may lead to
the same ‘voracity’ effect as was described in the previous section.
7. Conclusion
Some answers for the questions posed in the introduction can now be provided.
The waves of expropriations seen in the developed world in the 1960s and
1970s were a result of the high real prices for the minerals at the time, as well
as, to some extent, the wave of independence across developing countries. The
combination of high real prices and the low profit-sharing of the contracts at
the time gave the host governments large incentives to expropriate. Developed
countries are no different from developing countries in this regard, as Australia
and Canada had a similar history of expropriation during times of high prices.
The standard explanation for expropriation – that of political or economic
crises within the host country – is incorrect for expropriations within the
minerals sector. However, politics is an important explanatory factor in
expropriation as democracies were observed to be far more prone to expropriation than dictatorships.
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
100
R. Duncan
Facing uncertainty over future minerals prices, the best defence for foreign
investors against expropriation is to maintain a large domestic ownership/
profit share in the project. Such contracts lessen the gap between the value of
a firm in expropriation and the tax revenues of the domestic government
from the project, and thus the temptation to expropriate.
Returning to an era of depressed mineral prices and little expectation of
future price rises in the 1980s and 1990s, the decisions by those developing
countries that expropriated in the past now look terrible. Contracts based on
profit-sharing in an era of low profits generated little tax revenue, and nationally owned mines did not generate revenues needed for reinvestment.
This new depressed environment has generated a relatively recent response
by developing countries to divest themselves of minerals assets. The question
remains, however, about the behaviour of developing country governments if
the world moves into another period of rising mineral prices in the future.
The recent shift to democratic political institutions in many developing countries may cause problems in the future if mineral prices begin to rise.
References
Atkeson, A. (1991). International lending with moral hazard and risk of repudiation, Econometrica 59, 1069–1089.
Baltagi, B.H. (1995). Econometric Analysis of Panel Data. John Wiley & Sons, New York.
Banks, A., Muller, T. and Overstreet, W. (2003). Political Handbook of the World 2000–2002.
CSA Publications, Binghampton.
Barro, R.J. and Lee, J.-W. (1994). Sources of economic growth, Carnegie-Rochester Conference
Series on Public Policy 40, 1 – 46.
Basche, J.R. (1979). Nationalization: the experience of U.S. companies in the 1970s, The Conference Board Information Bulletin, no. 62, 1–18.
Beck, T., Clarke, G., Groff, A., Keefer, P. and Walsh, P. (2001). New tools in comparative political
economy: the database of political institutions, World Bank Economic Review 15, 165–176.
Bohn, H. and Deaton, R.T. (2000). Ownership risk, investment and the use of natural
resources, American Economic Review 90, 526–549.
Cole, H.L. and English, W.B. (1991). Expropriation and direct investment, Journal of International Economics 30, 201–227.
Eaton, J. and Gersovitz, M. (1984). A theory of expropriation and deviations from perfect capital mobility, Economic Journal 94, 16 – 40.
Garnaut, R. and Clunies Ross, A. (1975). Uncertainty, risk aversion and the taxing of natural
resource projects, Economic Journal 85, 272–287.
International Monetary Fund. (various years). International Financial Statistics. International
Monetary Fund, Washington, DC.
Jodice, D.A. (1980). Sources of change in Third World regimes for foreign direct investment,
1968–1976, International Organization 34, 177–206.
Jones, R.J. Jr (1980). A model for predicting expropriation in Latin America applied to
Jamaica, Columbia Journal of World Business 15, 74–80.
Jones, R.J. Jr (1984). Empirical models of political risk in U.S. oil production in Venezuela,
Journal of International Business Studies 15, 81–95.
Knudsen, H. (1974). Explaining the national propensity to expropriate: an ecological
approach, Journal of International Business Studies 5, 51–71.
Kobrin, S.J. (1979). Political risk: a review and reconsideration, Journal of International Business
Studies 10, 67–80.
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006
Price or politics
101
Kobrin, S.J. (1980). Foreign enterprise and forced divestment in LDCs, International Organization
34, 65–88.
Marshall, M. and Jaggers, G. (2001). Polity IV. Integrated Network for Societal Conflict
Research (INSCR) Program, Center for International Development and Conflict Management, University of Maryland, College Park.
Picht, H. and Stüven, V. (1991). Expropriation of foreign direct investments: empirical evidence
and implications for the debt crisis, Public Choice 69, 19–38.
Prain, R. (1975). Copper: The Anatomy of an Industry. Mining Journal Books, London.
Raff, H. (1992). A model of expropriation with asymmetric information, Journal of International Economics 33, 245–265.
Schnitzer, M. (1999). Expropriation and control rights: a dynamic model of foreign direct
investment, International Journal of Industrial Organization 17, 1113–1137.
Schnitzer, M. (2002). Debt v foreign direct investment: the impact of sovereign risk on the
structure of international capital flows, Economica 69, 41–67.
Singer, H.W. (1950). The distribution of gains between investing and borrowing countries,
American Economic Review: Papers and Proceedings 40, 473–485.
Thomas, J. and Worrall, T. (1994). Foreign direct investment and the risk of expropriation,
Review of Economic Studies 61, 81–108.
Tornell, A. and Lane, P.R. (1999). The voracity effect, American Economic Review 89, 22–46.
U.S. Bureau of Mines. (various years). Minerals Yearbook. US Department of Interior, US
Government Printing Office, Washington, DC.
U.S. Geological Survey. (2005). Historical statistics for mineral and material commodities in
the United States. Open-File Report 01-006, U.S. Department of the Interior. Available
from URL: http://minerals.usgs.gov/minerals/pubs/of01-006/ [accessed 22 September 2005].
United Nations Statistics Division. (2005). National Accounts Main Aggregates Database.
United Nations. Available from URL: http://unstats.un.org/unsd/snaama/ [accessed 26 September 2005].
Veugelers, R. (1993). Reputation as a mechanism alleviating opportunistic host government
behavior against MNEs, Journal of Industrial Economics 41, 1–17.
Williams, M.L. (1975). The extent and significance of the nationalization of foreign-owned
assets in developing countries, 1956–1974, Oxford Economic Papers 27, 260–273.
© Australian Agricultural and Resource Economics Society Inc. and Blackwell Publishing Ltd 2006